Related Experiment Video
Updated: Jun 19, 2026
08:45
Mapping Cortical Dynamics Using Simultaneous MEG/EEG and Anatomically-constrained Minimum-norm Estimates: an Auditory Attention Example
Published on: October 24, 2012
14.6K
Single-microphone deep envelope separation based auditory attention decoding for competing speech and music.
M Asjid Tanveer1, Jesper Jensen1,2, Zheng-Hua Tan1
1Department of Electronic systems, Aalborg University, Aalborg, Denmark.
Journal of Neural Engineering
|April 25, 2025
Summary
This study presents a deep learning system for separating audio sources and decoding auditory attention (AAD) using single-microphone electroencephalography (EEG) data. The system effectively decodes attended speech or music, even in complex acoustic environments.
Area of Science:
- Neuroscience
- Signal Processing
- Machine Learning
Background:
- Auditory attention decoding (AAD) aims to identify attended sound sources from neural signals.
- Single-microphone systems face challenges in separating competing audio sources like speech and music.
- Deep learning offers potential for robust source separation and AAD in complex auditory scenes.
Purpose of the Study:
- Introduce an end-to-end deep learning system for source separation and AAD using single-channel EEG.
- Evaluate the system's performance in distinguishing between speech and music targets and distractors.
- Assess the model's generalization capabilities across different acoustic conditions and head-related transfer functions (HRTFs).
Main Methods:
- A deep learning model was developed for source envelope separation directly from the mixed audio signal.
- Electroencephalography (EEG) signals were used for deep stimulus reconstruction, with Pearson correlation as the loss function.
- Models were trained and evaluated on speech/music pairs, incorporating 10 HRTF variants to simulate diverse head and ear effects.
Main Results:
- The system achieved a target accuracy of 82.4% on the original dataset and 75.4% across HRTF variants.
- Distractor audio achieved a low Pearson correlation of 0.004, indicating successful separation.
- AAD accuracy remained high across different speech and music combinations, with performance comparable to perfect source separation.
Conclusions:
- The developed deep learning models demonstrate strong generalization for source envelope separation and AAD across various speech, music, and HRTF conditions.
- While source separation is slightly less effective for mixed music and speech, it does not negatively impact AAD performance.
- The study validates the potential of single-microphone deep learning systems for robust auditory attention decoding.
Related Concept Videos
01:17Auditory Perception
The auditory system is essential for sound perception, utilizing various critical structures. When sound waves enter the outer ear, they travel through the ear canal and cause the eardrum to vibrate. These vibrations are then transmitted to the middle ear, where three tiny bones – the malleus, incus, and stapes – amplify the sound. This amplification is crucial, as it ensures that the sound vibrations are strong enough to be conveyed to the inner ear. These vibrations then reach the cochlea, a...
01:21Perceiving Loudness, Pitch, and Location
The human brain perceives pitch through two primary mechanisms reflected in place theory and frequency theory. Each mechanism describes how sound waves are interpreted as specific pitches by the brain, offering insights into the intricate processes of auditory perception.
Place theory, or place coding, suggests that different pitches are heard because various sound waves activate specific locations along the cochlea's basilar membrane. The brain determines the pitch of a sound by identifying...
Place theory, or place coding, suggests that different pitches are heard because various sound waves activate specific locations along the cochlea's basilar membrane. The brain determines the pitch of a sound by identifying...